Interpersonal Effects of Expressed Anger and Sorrow in Morally Charged Negotiation

Judgment and Decision Making, Vol. 9, No. 2, March 2014, pp. 104–113

Interpersonal effects of expressed anger and sorrow in morally charged negotiation

Morteza Dehghani∗ Peter J. Carnevale † Jonathan Gratch ‡

Abstract

The expression of emotion can play a significant role in strategic decision-making. In this study, we hypothesized that emotion expression alters behavior in morally charged negotiation. We investigated the impact of facial displays of discrete emotions, specifically anger and sadness, in a morally charged multi-issue negotiation task. Our results indicate that if a negotiator associated moral significance to the object of the negotiation, displays of anger resulted in reduced concession making whereas displays of sadness increased concession making. Moral significance of the issues fostered an emotional matching mechanism of sorrow, where a sorrow expression from one party elicited a sorrow expression from the other. Taken together, the results indicate that emotional expressions can affect morally charged negotiation in ways that can inhibit as well as promote cooperation. Keywords: emotion, sacred values, moral decision-making, negotiation.

1 Introduction cred domains. In this study, we investigated the impact of facial dis- Recent research into sacred values reveals a consistent pat- plays of discrete emotions, specifically sadness and anger, tern: when confronted with possible threats to moral or in a morally charged multi-issue negotiation task in which sacred concerns, people tend to react emotionally (e.g., one of the issues entailed the moral matter of saving the Ginges, Atran, Medin, & Shikaki, 2007), become uncom- life of the negotiator’s child. We suggest that interper- promising (e.g., Tetlock, 2003), and act in ways contrary sonal effects of emotion unfold differently in moral con- to traditional formalizations of rational self-interest (e.g., texts and with morally charged issues in negotiation. This Atran, 2010). This article offers some hope in this oth- idea stems from work showing that people who have a sa- erwise gloomy picture. Building on findings from both cred value tend to reject tradeoffs with other values (es- moral decision-making and the interpersonal effects of pecially with secular issues such as money) and express emotion, we show that emotion, in certain circumstances, anger when such tradeoffs arise (e.g., Tetlock, Kristel, El- may foster cooperation in value-laden conflicts. We show son, Green, & Lerner, 2000). that well-established interpersonal effects of emotion unfold differently in moral domains. For example, displays We based our predictions on the notion that anger con- of sadness can convey weakness (Tiedens, 2001; Tiedens, notes a violation of autonomous individual rights (Rozin, Ellsworth, & Mesquita, 2000) and promote exploitation in Lowery, Imada, & Haidt, 1999), whereas sadness connotes non-moral negotiations (Van Kleef, De Dreu, & Manstead, sympathy, need, weakness and concern (Horbeg, Oveis 2006). This effect reverses in strategic decision making & Keltner, 2011). We expected that when an object is over sacred issues. Our findings have potential important perceived as a sacred (or protected) value, with intrin- implications for negotiation and conflict resolution in sa- sic moral significance (Tetlock, 2003; Baron & Spranca, 1997), angry or sad facial displays expressed by an opponent will have opposing effects on the negotiator. Thus, This research was supported by AFOSR Young Investigator, Program we hypothesized that perceiving different emotional ex- AFOSR FA9550-09-1-0507, NSF IIS-1211064, and NSF SES-0836004. The software developed for the experiments reported here is free and pressions in others would influence moral cognition by available by contacting Morteza Dehghani at [email protected] shifting interpretive-frames. In other words, we suggest Copyright: © 2013. The authors license this article under the terms of that the interpretation of a moral issue can be affected by the Creative Commons Attribution 3.0 License. the emotional expressions conveyed by the opposing ne- ∗Brain and Creativity Institute, University of Southern California, 3620A McClintock Avenue; Los Angeles, CA 90089, USA Email: mde- gotiator in the negotiation, in addition to the emotion ex- [email protected] perienced by the negotiator, as shown in past work. † Marshall School of Business, University of Southern California, Specifically, we hypothesize that when facing an an- 3670 Trousdale Parkway, Los Angeles, CA 90089-0808, USA ‡Institute for Creative Technologies, University of Southern Califor- gry opponent, a negotiator dealing with a sacred-value nia, 1205 Waterfront Dr., Playa Vista, CA 90094-2536, USA issue—a morally significant issue—would show rejection

104 Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 105

of tradeoffs and concede very little, as their concerns about emotional display. For example, Van Kleef, De Dreu and their sacred values would be amplified by the anger ex- Manstead (2004) showed that negotiators conceded more pressed in the other party. That is, expressed anger would to an angry opponent than to a happy one and further pro- lessen the likelihood of concessions for sacred-value is- vided evidence that participants used expressed emotion sues. However, when interacting with an opponent who to infer the other party’s limits and adjusted their offers displays sadness, the morality of sympathy, need, weak- accordingly. This line of research argues that anger com- ness and concern become salient, and participants may municates that a party has high aspirations and that con- thus make greater concessions. For non-sacred-value par- cessions are thus required to reach an agreement. ticipants we expected to see the known pattern of in- Although less studied than other emotions, sadness is creased concession due to perceived anger, consistent with another emotion that naturally arises in intergroup con- findings of Van Kleef and others in non-moral domains flicts. Sadness can communicate weakness or low status (e.g., Van Kleef, et al., 2004; Sinaceur & Tiedens, 2006). (Tiedens, 2001; Tiedens, Ellsworth & Mesquita, 2000). Our hypothesize is built on Horberg, Oveis and Kelt- From a functional perspective, sadness can serve as an ap- ner’s (2011) theory that different moral concerns get pri- peal for support (Tomkins, 1963) that aims to elicit empa- oritized based on the distinct emotions that are experi- thy and helping behavior (Yee & Greenberg, 1998). How- enced. This theory, rooted in social-functional framework ever, there is no clear consensus on the impact of sadness of emotion (Keltner & Haidt, 1999; Frijda & Masquita, on negotiations. As a signal of weakness, one might ex- 1994), argues that our perception of the permissibility of pect sadness to provoke tough stances by the other party to actions in moral situations is affected by the emotions ex- strengthen their negotiation position (Tiedens, 2001; Van perienced. For example, disgust has been linked to viola- Kleef, De Dreu & Manstead, 2006). On the other hand, as tions of purity-sanctity (Rozin, et al., 1999); experimen- an elicitor of support and sympathy, sadness might trigger tally predisposing individuals to disgust increases their feelings of compassion and greater concessions. Few stud- tendency to focus on purity related issues (e.g., sexual- ies have examined this issue directly, although one line of ity) as opposed to other moral concerns such as justice studies by Van Kleef and colleagues (2006) suggests that (e.g., Tapias, Glaser, Keltner, Vasquez & Wickens, 2007). displays of disappointment (which in some contexts re- In a related vein, DeSteno, Dasgupta, Bartlett, and Cajdric lates to sadness) will tend to elicit greater concessions in (2004) report that the display of anger can evoke automatic non-sacred domains. bias directed toward out-groups. In contrast to the large volume of research into both However, the research on emotion-related moral ap- moral-decision making and the interpersonal function of praisals has been for the most part limited to the intrap- emotions in negotiations (mainly focusing on issue that ersonal effects of emotion in decision-making. In con- may be of interest to people but have no sentimental or trast, research on negotiation and conflict resolution has moral significance to them), almost no studies exist in the considered both the intrapersonal as well as the interper- intersection of these two fields. Yet several new findings sonal effects of emotion (e.g., Carnevale, 2008; Carnevale hint that studying this intersection may provide new in- & Isen, 1985; O’Quin & Aronoff, 1981). Carnevale and sights into the mechanisms underlying emotion’s impact Isen (1985) found that positive mood had an impact on on social behavior. For example, Harinck and Van Kleef negotiation, that is, it increased problem solving and co- (2012) found that an expression of anger can have opposite operation, but not when a physical barrier separated the effects on reactions to an authority who denied a request negotiators so that they could hear but not see one another, depending on the value-context of the request.1 Before thus preventing them from seeing one another’s facial ex- 1 pressions. This suggests that intrapersonal emotion may Harinck and Van Kleef (2012) found that the expression of anger had opposite effects on reactions to an authority who denied a request de- influence interpersonal emotion via facial expression or pending on the value-context of the request. They had participants play some other visually signaled expression of emotion. the role of a “Junior Trainee” who made a request to an “HR employee,” The interpersonal effect of emotion, for example, the in a hypothetical scenario, for either “30 days vacation” or “30 days off to take care of mother [who recently broke her leg].” The HR person re- impact of an adversary’s facial expression of emotion, sponded either “This request makes me really angry. . . take 15 days. . . ” might reflect in part appraisal theory mechanisms of emo- or was neutral (identical text minus the anger statement). When the con- tion: that is, emotional expressions by one party can pro- text was self-interest only (no mother involved), the participants reacted vide information about how they construe the situation. no differently to the anger response and the neutral response. However, when the context had care-value, participants indicated that they would This can influence the other party by shaping how they, escalate the matter (e.g., “I would contact my boss’s supervisor”) more in turn, construe the situation from their own perspec- when there was anger expressed than when no anger. Although the Har- tive, and thereby influence their emotions and decisions inck study is interesting in showing a boundary condition of the impact (Ames & Johar, 2009; de Melo, Carnevale, & Gratch, of anger, there are important differences in method, results, and interpre- tations between that study and the present study that are worth noting. 2014; Sinaceur & Tiedens, 2006; Van Kleef, De Dreu & First, the domain was not negotiation but was reactions to an authority Manstead, 2010). Several studies indicate the impact of an that made a decision. This might explain why the often-found effect in Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 106

different objects placed on a board and the participant and Figure 1: Screenshot of the board where negotiation takes the agent take turns in taking ownership of some of the ob- place. jects and giving others away (Figure 1). Participants move items around the board by grabbing them with a mouse and cursor and placing the items either on their own side or on the opponent’s side. After each offer by the participant, the agent evaluates the offer, expresses an emotional reaction to the offer and decides whether to accept the offer or propose a new offer. Participants can express emotional reactions at any point by choosing one of the emotional facial displays at the bottom right corner of the screen (Figure 1). This task is an objects-based form of the more abstract, computerized, negotiation task common in the negotiation literature and used in past studies of negotiation and emotion expression (e.g., Hilty & Carnevale, 1993). Aspects of this task are easily configurable for a variety of experiment questions. In our case, all items appeared in the middle section of the board and were up for grabs. The discussing our experiments and results, we first describe negotiation consisted of 6 rounds with each player taking the negotiation task that we used. turns making or receiving offers six times. When a participant made an offer, the computer opponent decided to accept or reject the offer based on a pre-programmed strat- 2 Objects Negotiation Task egy unknown to the participant. If the opponent decided to reject the offer, it made a new proposal that the participant Medin and Atran (2004) argued that research on morally could in turn accept or reject. motivated decision-making relies heavily on a “narrow empirical base” in regards to the subject populations as well as to the stimuli used in the experiments. Samples 2.1 Agent Offers used in these studies mostly comprise undergraduates at All agents in this study followed the same strategy—a major research universities; the scenarios and the stim- fixed, non-contingent series of offers that was designed uli materials used mainly focus on single-shot trade-offs to simulate resistance to making concessions. In pilot scenarios in which participants judge the permissibility of testing, participants perceived this policy to be tough but set hypothetical actions (e.g., killing one person instead plausible. There are four different groups of items in- of five). However, many real-life moral situations un- volved in the negotiations (medicine, water bottles, food fold over repeated interactions, sometimes spanning years, cans, money), with three items per group. The negotiation such as intractable socio-political conflicts involving sa- strategy of the agent is as follows ([medicine, water, food, cred values, e.g., the Israel-Palestine conflict (Ginges et money]): Round 2: [0, 0, 0, 0]; Round 4: [0, 0, 0, 1]; al., 2007) and the Iran nuclear conflict (Dehghani, et al., Round 6: [0, 0, 0, 1]; Round 8: [0, 1, 0, 1]; Round 10: 2009, 2010). [0, 1, 0, 1]; Round 12: [0, 2, 0, 2], where the numbers in To overcome some of the above shortcomings, we de- the brackets represents how many items in each group the veloped a web-based multi-round negotiation task involv- agent chooses to give to the participant. In the decision- ing a participant and an opponent (computer-agent), with making algorithm of the agent, the items are given the fol- the negation literature of anger enhancing concession making did not oc- lowing qualitative payoff values: [50, 10, 5, 1]. These cur in the Harinck study. Second, the request from a “Junior Trainee” payoff values are only used for internal calculations and for a 30 day vacation may have been seen as inappropriate; thus, the less harsh reaction to denial of the request in the no-mother condition. Third, are not shown to the participants. The agent accepted a the anger expression was text; the present study used an anger facial ex- participant’s offer if it had a higher or equal overall utility pression; there may be similarities and differences between the two, in than the offer that the agent was about to make. Otherwise, particular with regard to reciprocity of emotion expression and apparent it rejected it and made its next offer. sincerity. Fourth, unlike Harinck, the present study used a comparison to another emotion, sorrow, which has a basis in moral compassion, and for which we expected, and found, an effect on negotiation that is distinct from the effect of anger. Hence, we see some relevance, but also some 2.2 Agent Expressions disconnect, between the Harinck study and the present study, in particular in terms of its connection to the dynamics of achieving agreement in In the first experiment, agents followed one of three pos- negotiations that involve sacred values. sible facial display policies depending on the condition. Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 107

Figure 2: Facial displays used in the experiments (Anger, Figure 3: In Experiment 1, participants were presented Neutral, Sorrow). either with scenario A (deadly-infection) or scenario B (minor-cold). Imagine the following hypothetical scenario: There has been an earthquake in the town you live in and many have been injured. All roads to your town have been blocked and as a result aid is coming in very slowly. Because of this every family has to split packages of aid sent using he- licopters with another family. The angry agent followed a fixed, non-contingent policy, You and the family that have to split the aids with each other, displaying anger on rounds 2, 6 and 10, and returned to a both have babies who have [A: been injured and have devel- neutral face after five seconds (i.e., the policy is the same oped infections] [B: have caught minor colds]. [A: The only no matter what the participant offered). The sad agent fol- way to control the spread of infection, which if not stopped lowed the same policy but displayed sadness rather than will become lethal, is to use penicillin] [B: In order to relieve anger. In all other rounds, both agents displayed a neutral the cold you can give your child acetaminophen]. You are also face. In the second experiment, in addition to sadness and running low on food, but have enough clean water that would last you for several days. All the shops in the town are closed, anger we had a neutral condition in which the agent dis- so it is uncertain whether you can use the money to purchase played a neutral face throughout the negotiation. Figure 2 goods. shows the expressions2. Given the circumstances, you know that no other aid package will be received for another week. The aid packages include medicine including [A: penicillin] [B: acetaminophen], canned 3 Experiment 1 food, some money and water bottles. In the task that follows, you have to negotiate how these items In this experiment, participants engaged the Objects Ne- have to be split between your family and the other household. gotiation Task with an opponent that followed either an- You do not know how much food and water the other family gry or sad facial display policies. We predicted that par- has. ticipants who view an object of the negotiation as sacred The negotiation is done in a sequence of alternating offers. would show the typical rejection of tradeoffs and concede You will make the first offer. The other negotiator may or may less, as seen in past work (e.g., Tetlock, et al., 2000), but not accept your offer. If it does not accept it, that is, if it rejects only when the opponent displayed anger. However, when your offer, it will send you a new offer. You can either accept participants with a sacred value face an opponent who dis- or reject its offer. If you accept it, you will get to keep the played sorrow, we expected them to concede more. items that you did not give them. If you reject their offer, you can make another offer and submit it to them. If after 6 rounds there is no agreement, the negotiation will end in no agreement. 3.1 Method In this case, you both will only receive one of each item and the rest will be given away. 3.1.1 Participants Try to get as many items as you can. 332 American Amazon-Turk workers (age: 33.27, gender: In the task, to review, you will negotiate with the other family 51% female) were paid $1 each to participate in our study. over the aid packages that include: On average it took each participant 5 minutes and 49 seconds to complete our task. 1. Medicine (penicillin):

3.1.2 Design and procedure The study employed a between subject 2 X 2 X 2 full fac- 2. Water Bottles: torial design, where the ﬁrst factor was agent’s expressed emotion (anger/sadness), the second factor was the ex- 3. Money: perimental scenario (deadly-infection scenario/minor-cold scenario, Figure 3), and the last factor was whether or not 4. Canned Food: participants held a sacred value for the medicine package. The main dependent variable in our experiment was the amount of concession in the negotiation (demand at round 2These drawings have been used with the artist’s permission. Please visit http://www.scottmccloud.com/ for more information about his one subtracted from demand at the last round of negotia- work. tion). Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 108

Participants first received one of the two scenarios de- Figure 4: Concessions for medicine as a function of scribed in Figure 3. In the deadly-infection scenario, the scenario stated that both their child and the other fam- Agents’ emotional expressions and Sacred Value (SV). ily’s child were injured and developed infections and the The range for the Y-axis is from 3 to -3. only way to control the spread of infections, which if not Sacred Value stopped would become lethal, was to use penicillin in- NoSV cluded in the medicine package. In the minor-cold sce- SV nario, the scenario stated that both their child and the other family’s child had caught minor colds and to relieve the 0.6 cold they could give their child acetaminophen included in the medicine package. Both scenarios stated that they were running low on food, but had enough clean water that would last several days. They further learned that all 0.4 shops in the town closed, so uncertain whether they could use any money to purchase goods. After reading one of the scenarios, we assessed partici- Concessions pants’ values regarding the medicine package using Baron 0.2 and Spranca’s (1997) measure of sacred value. In line with this measure, we asked participants "How do you feel about giving up the medicine package?" and they received the following four choices: 0.0 a. I think this definitely needs to happen. b. I do not object to this. c. This is acceptable only if the benefits of trading the EveryOtherTurnAngry EveryOtherTurnSad Agent Expressed Emotion medicine are great enough. d. This should not be done no matter how great the benefits. 2 emotion F (1, 292) = 8.774, p = .003, ηpartial = 0.029 We defined a participant who selected “d” as having a (Figure 4). As expected, anger (M = 0.478) resulted in sacred value for the medicine package. Participants then higher concessions, compared to sadness (M = 0.240) for non-sacred-value participants t(165) = 1.905, p = 0.058, played the Objects Negotiation Task, in this case with one 2 of the objects was construed to be sacred. r = 0.021. However, sacred value participants who interacted with the sad agent (M = 0.403) conceded significantly higher than sacred value participants who interacted 3.2 Results with the angry agent (M = 0.033) t(131) = 2.576, p = .011, 2 Participants who dropped out of the negotiation before r = 0.048. Interestingly, sacred value participants who in- Round 3 (made only one or two offers and were exposed teracted with the angry agent (M = 0.033) conceded much to the emotional displays of the agent only once) were ex- less on medicine than non-sacred-value participants interacting with the same agent (M = 0.478) t(151) = 3.2067, cluded from the analysis (N = 32, 9.6% of the sample). 2 44% of the participants reported having a sacred value for p = .0016, r = 0.064. medicine. We also analyzed the participants’ expressed emotion We first analyzed the differences in concession rates be- throughout the length of the negotiation. There was a tween the groups. We used demand difference (number of significant difference in expressed anger between sacred packages demanded in the first offer deducted from last value participants who interacted with the angry agent offer) for medicine as a dependent variable in a 2 X 2 X (M = 1.623) and non-sacred-value participants interacting 2 ANOVA where the first factor was the displayed emo- with the same agent (M = 1.143) with sacred value par- tional reaction (sadness/anger), the second factor was the ticipants expressing more anger t(151) = 2.234, p = .027, presence or absence of sacred values and the third factor r2 = 0.032. Also, sacred value participants who interacted was scenario (deadly infection vs. minor cold). The sce- with the agent that displayed sad facial expressions (M = nario manipulation did not have an effect on the course of 0.875) expressed more sadness than sacred value partic- the negotiation (p = 0.824). There was a significant in- ipants who interacted with the angry agent (M = 0.541) teraction between sacred value and the agent’s displayed t(131) = 1.934, p = .055, r2 = 0.028. Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 109

3.3 Discussion Figure 5: Scenario used in Experiment 2. The data from Experiment 1 indicated that, for the Imagine the following hypothetical scenario: medicine package, the amount of concession made by par- There has been an earthquake in the town you live in and ticipants depended on the agent’s displayed emotion and many have been injured. All roads to your town have been whether they viewed the item as sacred. As predicted, blocked and as a result aid is coming in very slowly. Because sacred value participants who interacted with the agent of this every family has to split packages of aid sent using he- that displayed anger made significantly smaller conces- licopters with another family. sions compared to sacred value participants who interacted The aid packages that have arrived include: 1. Medicine with the sad agent. (penicillin), 2. Water bottles, 3. Some money and 4. Canned food. In the task that follows, you will negotiate how these will be split between your family and the other family. 4 Experiment 2 It is important to realize that given the circumstances, no other aid package will be received for another week. And all We designed Experiment 2 to address several limitations the shops in town are closed, so it is doubtful that money can to the first experiment and to shed light on the mechanism be used to purchase goods. associated with the effects. First, Experiment 1 lacked a You are running low on food and clean water, so this aid is neutral-emotion control condition, meaning that we did valuable. You do not generally know about the situation that not know whether expressed anger or sorrow, or both, the other family is in. were critical for the impact of emotion on negotiation. However, there are two important things that you do Therefore, in Experiment 2, in addition to anger and sad- know for sure: 1. You need penicillin for your family, and ness, we investigated the effects of an emotionally neutral 2. You have just learned that the other family needs peni- agent on the negotiation. cillin. Both families have a baby, and both babies have been Second, given that in Experiment 1 we asked the sa- injured and have developed serious infections. The only cred value question only about the medicine package, we way to control the spread of these infections, which if not stopped will cause their baby and your baby’s death, is to could not investigate the interplay between emotions and use penicillin. You, and they, both need a lot of penicillin. possible sacred values for other items. In Experiment 2, In the task, to review, you will negotiate with the other family we asked the sacred value question for each item. This ad- over the aid packages that include: ditionally provided the opportunity to test our assumption that medicine to alleviate illness is a sacred value more than other issues. Some issues in negotiation, for exam- 1. Medicine (penicillin): ple, negotiations over a human life, or a family heirloom, are often seen sacred (Atran & Axelrod, 2008). Of course, the sacredness of an issue in negotiation is associated with 2. Water Bottles: the general value placed on the issue at the outset of negotiation. Indeed, at the outset of negotiation, a sacred issue 3. Money: can be defined as one in which the negotiator refuses to even consider being an issue that can be conceded in negotiation. In the present study, this is how we defined a 4. Canned Food: sacred value issue, following the procedure developed by Baron and Spranca (1997) by assessing the negotiators’ of exchange should not rise to the level of a sacred value. likelihood of concession on the issue at the start of nego- Evidence for this comes from work on the value of tokens, tiation and agreement with the statement “[. . . concession held for making trades, which do not show an endowment on this issue] shouldn’t be done no matter how great the effect (Kahneman, Knetsch, & Thaler, 1990). Thus we benefits.” This fits with our view that issues in negotia- expected negotiators to rate money as least sacred. tion that relate to matters of life and death are particularly In this experiment, we also assessed differences in sacred. Indeed, the recent US national poll conducted by moral concerns with the Moral Foundation Questionnaire Gallup, Inc. revealed that of the moral issues that drive (Haidt, Graham & Joseph, 2009) to investigate whether in- disagreement in the US, many are about creation of hu- teracting with different agents can shift participants’ moral man life or its termination, for example, cloning, abortion, concerns. We expected that the angry or sad emotional the death penalty (Saad, 2010). expressions displayed by the other during the negotiation Thus, we expected medicine to save a life, as an issue, might have an impact on participants’ moral appraisals. to be sacred in the negotiations, more so than the other Specifically, we predicated that, for participants with sa- issues in the negotiation. Moreover, one issue that should cred values, display of sadness in others might increase typically not be sacred is money. Money held for purposes their sensitivity towards fairness. Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 110

4.1 Method items categorized as sacred values and one for those categorized as non-sacred). For example, if a participant rates 4.1.1 Participants medicine and food as sacred, the average demand differ- Seventy-two University of California at Los Angeles stu- ence for these two items and the average demand differ- dents (age=20.82, gender: 36.11% female) were recruited ence for money and water would be used as DVs cor- through the California Social Science Experimental Labo- responding to the two levels of the within-subject factor. ratory at UCLA. Each participant received $15 for partic- There were a number of missing data points, and we ex- ipating in this experiment. cluded them case-wise for this analysis. As predicted, there was a significant interaction between sacred values and displayed-emotion F (2, 36) = 2 4.1.2 Design 3.400, p = 0.044, ηpartial = 0.075. Similar to previous findings, display of anger resulted in higher concessions This experiment employed a mixed-design with sacred for non-sacred values (M = 0.432) compared to expression values as the within subject factor and the agent’s ex- of neutral emotion (M = 0.127) t(38) = 2.011, p = 0.051, pressed emotion (anger/neutral/sadness) as the between r2 = 0.096. However, display of anger for sacred values subject factor. The experimental scenario used in this ex- (M = 0.166) resulted in lower concession compared to sad- periment was a modified version of the deadly-infection ness (M = 0.643) t(24) = 2.112, p = 0.045, r2 = 0.157, but scenario (Figure 5). After reading the scenario, we as- not compared to neutral expressions (M = 0.243) t(23) = sessed participants’ values regarding all the packages us- 0.408, p = n.s.. Replicating the results of the first exper- ing Baron and Spranca’s (1997) measure described in Ex- iment, participants with one or more sacred values who periment 1. After the negotiation, the participant filled out interacted with the sad agent (M = 0.643) had higher con- the Moral Foundations Questionnaire (Haidt, Graham & cessions compared to those without a sacred value in the Joseph, 2009). same condition (M = 0.321, tpaired (13) = 1.945, pone-tailed = 0.037), and also to participants in the neutral conditions 4.2 Results (sacred value neutral: M = 0.243, t(25) = 2.012, p = 0.055; non-sacred value/neutral: M = 0.127, t(31) = 2.843, p = Similar to Experiment 1, participants who dropped out of 0.008), but not compared non-sacred values in the anger the negotiation before Round 3 were excluded from the condition (M = 0.443), t(85) = 1.39, p = n.s. (Figure 6). analysis (N = 8). One additional subject was excluded We also analyzed the participants’ expressed emotions. because the subject’s demand difference (DV) was over There was no within-subject factor for this analysis; we twice the standard deviation from the mean. used sacred value for medicine and agent’s expressed The data on the ratings of the four issues for sacredness emotion as between subject factors. A 3 (agent’s ex- (the Baron and Spranca 4-level measure) indicated that, pressed emotion: sadness/neutral/anger) X 2 (sacred value as expected, none of the participants (0%) rated money / no sacred value for medicine) ANOVA with frequency of to be a sacred value. However, most saw the medicine participants expressed sadness as dependent variable re- package as a sacred value (57.14%), and this is a signif- vealed a significant interaction between agent’s expressed icantly higher level than the other three items (compared emotions and sacred value F (2, 57) = 4.043, p = 0.023, 2 2 2 to money: χ = 47.640, p < .001; food: χ = 30.036, p < ηpartial = 0.500. Participants who did not indicate a sacred .001; water: χ2 =16.163, p < .001). The other negotiation value for medicine and interacted with the angry agent (M items were less likely rated sacred (for the water package: = 1.778) expressed sadness more frequently than partici- 19.05%; for the food package, 7.93%). These rating lev- pants who indicated a sacred value for medicine and in- els were different from money (water: χ2 = 12.351, p < teracted with the same agent (M = 0.750) t(19) = 2.555, 0.001; food: χ2 = 4.375, p = 0.036) but not statistically p = 0.019, r2 = 0.255. More importantly, replicating the different from each other. 60.56% of participants rated at findings of the previous experiment, participants who in- least one item as a sacred value. dicated a sacred value for medicine and interacted with the Unlike Experiment 1 where we categorized sacredness sad agent (M = 1.571) expressed more sadness than partic- based on the sacred value measure for medicine, in this ipants who held medicine as a sacred value but interacted experiment we investigate the interplay between emotions with the angry agent (M = 0.750) t(24) = 1.763, pone-tailed 2 and all items categorized as sacred value. The data were = 0.045, r = 0.115. There were no reliable differences in analyzed using a mixed-design ANOVA with one within- the frequency of expressed anger (Figure 7). subject factor (sacred value /no sacred value), and one be- We then analyzed participants’ answers to the Moral tween subject factor, the agent’s expressed emotion (sad- Foundations Questionnaire. There were not any signif- ness/neutral/anger). The DV used in this analysis was icant differences across conditions, except for a non- the mean number of conceded items (i.e., one mean for significant trend where participants who interacted with Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 111

the sad agent (M = 4.608) expressed higher concern in the Figure 6: Concessions as a function of Agents’ emotional moral domain of fairness compared to the sacred value expressions and Sacred Value (SV). The range for the Y- participants who interacted with the angry agent (M = axis is from 3 to -3. 4.264). 1.00 Sacred Values NonSVItemMean SVItemMean 4.3 Discussion

In this experiment, we replicated and extended the re- 0.75 sults of Experiment 1. Consistent with findings of Van Kleef and others (e.g., Van Kleef et al., 2004; Sinaceur & Tiedens, 2006) in non-moral domains, anger expressions for Non-sacred value participants resulted in higher con- 0.50 cession rates. However, similar to the first experiment, sacred value participants interacting with the angry agent showed the typical rejection of tradeoffs, while expression Concessions of sadness resulted in significantly higher concessions. Of 0.25 note, Experiment 2 found that none of the participants ranked money as a sacred value. This suggests that sorrow displays may not be effective in negotiation over money matters. 0.00 Moreover, in both experiments we found that sacred value participants who interacted with the sad agent expressed more sadness themselves, whereas non-sacred everyOtherTurnAngry alwaysNeutral everyOtherTurnSad value participants who interacted with the angry agent Agent Expressed Emotion expressed more sadness than participants who interacted with the sad agent. This is an indication that expression of sadness is interpreted differently depending on whether or Figure 7: Frequency of expressed sadness as a function of not the items of the negotiation are valued as sacred. Agents’ emotional expressions and Sacred Value (SV). We had predicted that witnessing sad facial displays would affect the decision-making of sacred value partic- 3 Sacred Values NoSV ipants by heightening the salience fairness (Horbeg, Oveis SV & Keltner, 2011). However, more explicit measures of fairness are needed to validate this. We speculate that the MFQ might not be a suitable measure for capturing ephemeral effects because of its length (32 questions).

2 5 General Discussion

Overall, the contribution of our work is two-fold. First, our results are consistent with the general view that social 1 context and emotion expression interact in their impact on behavior (Elfenbein, 2007). Second, we showed that expressing anger may not be the best strategy to achieve Frequency of Expressed Sadness higher concession rates in negotiation, a result that com- plements other studies that also show boundary conditions of the effects of anger displays in negotiation (Côté et al., 0 2013; Sinaceur, Adam, Van Kleen & Galinsky, 2013; Van Dijk, Van Kleef, Steinel, & Van Beest, 2008; Van Kleef everyOtherTurnAngry alwaysNeutral everyOtherTurnSad Agent Expressed Emotion & Côté, 2007). For example, anger expressions have been shown to be more successful in getting concessions when the recipient has poor alternatives (Sinaceur & Tiedens, 2006), when anger expressions are perceived as threats Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 112

(Sinaceur et al., 2011), when the recipient of the expres- References sions is in a low-power position, or when the expression of anger is perceived as appropriate by a high-power ne- Ames, D. R., & Johar, G. V. (2009). I’ll know what you’re gotiator (Van Kleef & Côté, 2007). Moreover, anger has like when I see how you feel: How and when affective been shown to promote conflict resolution (e.g. Halperin, displays influence behavior-based impressions. Psycho- 2011). Our result demonstrates that displays of anger may logical Science, 20(5), 586–593. result in lower concessions if one of the parties associates Atran, S. (2010). Talking to the Enemy: Violent Extrem- moral significance to the objects of the negotiation. With ism, Sacred Values, and What it Means to be Human. moral significance, a sorrow display may be a more effec- London: Penguin. tive strategy to elicit concessions. Our results complement Atran, S., & Axelrod, R. (2008). Reframing sacred values. findings about the role of symbolic offers in tradeoffs in- Negotiation Journal, 24(3), 221–246. volving sacred values (Ginges, et al. 2007, Atran & Ginges Baron, J., & Spranca, M. (1997). Protected values. Or- 2012). We speculate that symbolic offers produce conces- ganizational Behavior and Human Decision Processes, sion because they express sympathy. 70(1), 1–16. The data also suggest an interesting link between sa- Carnevale, P.J. (2005). Psychological barriers to negoti- cred values and the reciprocation of emotion in negotia- ation. In Shimon Shamir & Bruce Maddy-Weitzman tion: sacred value participants who interacted with the an- (Eds.), The Camp David Summit: What went wrong? gry agent expressed more anger than did non-sacred value (pp. 210–218). Brighton: Sussex Academic Press. participants interacting with the same agent. And the sa- Carnevale, P.J. (2008). Positive affect and decision frame Group Decision and Negotiation, 17 cred value participants who interacted with the agent that in negotiation. , 51–63. displayed sad facial expressions expressed more sadness. Carnevale, P.J., & Isen, A.M. (1986). The influence of In Experiment 2, non-sacred value participants who inter- positive affect and visual access on the discovery of in- acted with the angry agent were more likely to mismatch tegrative solutions in bilateral negotiation. Organiza- emotion, that is, to express sadness. tional Behavior and Human Decision Processes, 37, 1– An interesting question for future research might con- 13. sider the mechanism associated with sacred values in ne- Côté, S., Hideg, I., & Van Kleef, G. A. (2013). The con- gotiation that lead negotiators to reciprocate the other’s sequences of faking anger in negotiations. Journal of emotion, and the interesting possibility that interfering Experimental Social Psychology. 49, 453–463. with that mechanism, in the case of anger with the sacred Dehghani, M., Atran, S., Iliev, R., Sachdeva, S., Medin, value, might foster cooperation. This is akin to the “bar- D. & Ginges, J. (2010). Sacred values and conflict over rier effect” in negotiation, which is well known by pro- Iran’s nuclear program. Judgment and Decision Mak- fessional mediators, where separating the parties in hos- ing, 5, 540–546. tile contexts can facilitate cooperation (Carnevale & Isen, Dehghani, M., Iliev, R., Sachdeva, S., Atran, S., Ginges, 1986). Also, it would be interesting to investigate changes J. & Medin, D. (2009). Emerging sacred values: Iran’s in saliency of different moral concerns based on expe- nuclear program. Judgment and Decision Making, 4, rienced interpersonal emotions using more explicit mea- 930–933. sures of morality. de Melo, C., Carnevale, P.J., & Gratch, J. (2014). Read- Sacred values play important roles in many cultural and ing people’s minds from emotion expressions in inter- political conflicts (e.g., Ginges et al., 2007; Dehghani et dependent decision making. Journal of Personality and al., 2009, 2010; Sheikh, Ginges & Atran, 2013). We sug- Social Psychology, 106(1), 73–88. gest that moral concerns can shift within negotiation in DeSteno, D., Dasgupta, N., Bartlett, M. Y., & Cajdric, A. ways that promote cooperation and concession making. It (2004). Prejudice from thin air: The effect of emotion is interesting to speculate that U.S. President Jimmy Carter on automatic intergroup attitudes. Psychological Sci- orchestrated this sorrow-matching effect when he brought ence, 15, 319–324. Israeli President Begin and Egyptian President Sadat the Elfenbein, H. A. (2007). Emotions in organizations. The Gettysburg Battle Field Cemetery during the 1978 Camp Academy of Management Annals, 1, 315–386. David negotiations, which resulted in considerable con- Frijda, N. H., & Mesquita, B. (1994). The social roles and cession making (Carnevale, 2005). One real-world im- functions of emotions. In S. Kitayama & H. R. Markus plication of this research is that, in negotiation involving (Eds.), Emotion and culture: Empirical studies of mu- sacred values, displaying anger might lead to anger recip- tual influence. Washington, DC: American Psychologi- rocation, and lead conflicts to escalate. With sacred val- cal Association. ues, sympathy signals may result in better negotiation out- Ginges, J., Atran, S., Medin, D., Shikaki, K. (2007). Sa- comes. cred bounds on rational resolution of violent political Judgment and Decision Making, Vol. 9, No. 2, March 2014 Interpersonal effects of expressed anger and sorrow 113

conflict. Proceedings of the National Academy of Sci- fective in negotiations. Journal of Experimental Social ence, 104, 7357. Psychology, 42(3), 314–322. Haidt, J., Graham, J., & Joseph, C. (2009). Above and be- Tapias, P. M., Glaser, J., Keltner, D., Vasquez, K., & Wick- low left–right: Ideological narratives and moral founda- ens, T. (2007). Emotions and prejudice: Specific emotions. Psychological Inquiry, 20(2–3), 110–119. tions toward outgroups. Group Processes & Intergroup Halperin, E. (2011). Emotional barriers to peace: Neg- Relations, 10, 27–39. ative emotions and public opinion of Jewish Israelis Tetlock, P. E. (2003). Thinking the unthinkable: sacred about the peace process in the Middle East. Peace and values and taboo cognitions. Topics in Cognitive Sci- Conflict: Journal of Peace Psychology, 17, 22–45. ence, 7, 32. Harinck, F., & Van Kleef, G. A. (2012). Be hard on the in- Tetlock, P. E., Kristel, O., Elson, B., Green, M., & Lerner, terests and soft on the values: Conflict issue moderates J. (2000). The psychology of the unthinkable: Taboo the effects of anger in negotiations. British Journal of trade-offs, forbidden base rates, and heretical counter- Social Psychology, 51(4), 741–752. factuals. Journal of Personality and Social Psychology, Hilty, J., & Carnevale, P. J. (1993). Black-hat/white-hat 78, 853–870. strategy in bilateral negotiation. Organizational Behav- Tiedens, L. Z., Ellsworth, P. C., & Mesquita, B. (2000). ior and Human Decision Processes, 55, 444–469. Stereotypes about sentiments and status: Emotional Horberg, E. J., Oveis, C., & Keltner, D. (2011). Emotions expectations for high- and low-status group members. as moral amplifiers: An appraisal tendency approach Personality and Social Psychology Bulletin, 26, 560– to the influences of distinct emotions upon moral judg- 574. ment. Emotion Review, 3, 237–244. Tiedens, L. Z. (2001). Anger and advancement versus sad- Kahneman, D., Knetsch, J., & Thaler, R. (1990). Experi- ness and subjugation: The effect of negative emotion mental tests of the endowment effect and the Coase the- expressions on social status conferral. Journal of Per- orem. Journal of Political Economy, 98, 1325–1348. sonality and Social Psychology, 80, 86–94. Keltner, D., & Haidt, J. (1999). Social functions of emo- Tomkins, S. S. (1963). Affect, imagery, consciousness: Vol tions at four levels of analysis. Cognition and Emotion, 2. The negative affect. New York: Springer Verlag. 13, 505–521. Van Dijk, E., Van Kleef, G. A., Steinel, W., & Van Beest, Medin, D. L., & Atran, S. (2004). The native mind: Bi- I. (2008). A social functional approach to emotions in ological categorization and reasoning in development bargaining: When communicating anger pays and when and across cultures. Psychological Review, 111, 960– it backfires. Journal of Personality and Social Psychol- 983. ogy, 94, 600–614. O’Quin, K., & Aronoff, J. (1981). Humor as a technique Van Kleef, G. A., & Côté, S. (2007). Expressing anger in of social influence. Social Psychology Quarterly, 349- conflict: When it helps and when it hurts. Journal of 357. Applied Psychology, 92, 1557–1569. Rozin, P., Lowery, L., Imada, S., & Haidt, J. (1999). The Van Kleef, G. A., De Dreu, C. K. W., & Manstead, A. S. cad triad hypothesis: A mapping between three moral R. (2004). The interpersonal effects of anger and happi- emotions (contempt, anger, disgust) and three moral ness in negotiations. Journal of Personality and Social codes (community, autonomy, divinity). Journal of Per- Psychology, 86, 57–76. sonality and Social Psychology, 76, 574–586. Van Kleef, G. A., De Dreu, C. K. W., & Manstead, Saad, L. (2010). In 2010, Conservatives still outnumber A. S. R. (2006). Supplication and appeasement in moderates, liberals. Gallup. http://www. gallup. con[FB02?]ict and negotiation: The interpersonal ef- com/poll/141032/2010-Conservatives-Outnumber- fects of disappointment, worry, guilt, and regret. Jour- Moderates-Liberals. nal of Personality and Social Psychology, 91, 124–142 Sheikh, H., Ginges, J., & Atran, S. (2013). Sacred val- Van Kleef, G. A., De Dreu, C. W., & Manstead, A. S. R. ues in the Israeli–Palestinian conflict: resistance to so- (2010). An interpersonal approach to emotion in social cial influence, temporal discounting, and exit strategies. decision making: The emotions as social information Annals of the New York Academy of Sciences, 1299(1), model. In M. P. Zanna (Ed.), Advances in Experimen- 11–24. tal Social Psychology (Vol. 42, pp. 45–96). Burlington, Sinaceur, M., Adam, H., Van Kleef, G. A. & Galinsky, VT: Academic Press. A. D. (2013). The advantages of being unpredictable: Yee, J. L., & Greenberg, M. S. (1998). Reactions to crime How emotional inconsistency extracts concessions in victims: Effects of victims’ emotional state and type of negotiation. Journal of Experimental Social Psychol- friendship. Journal of Social and Clinical Psychology, ogy, 49(3), 498–508. 17, 209 –226. Sinaceur, M., & Tiedens, L. Z. (2006). Get mad and get more than even: When and why anger expression is ef-